leiten-zustand
v1.7.2
Published
Zustand state manager controllers
Downloads
380
Maintainers
Readme
Zustand controllers
Helps to avoid using other state managers to execute requests and allows you to work efficiently with zustand.
The Zustand Fetching Helpers library provides a set of functions and controllers that facilitate working with Zustand, a state management library. The functions described below are well-typed and allow working with nested objects. While Zustand suggests writing custom slices to divide the store into several parts, this library aims to simplify common data-related tasks without the need for additional state management solutions.
To get a better understanding of what this library offers and how it works, you can refer to the live example on CodeSandbox . In many cases, the provided controllers will help reduce the complexity of your store, eliminating the need to split it into multiple parts.
Installation
You can install the library using npm:
npm install leiten-zustand
Since "Zustand" translates to "state" in German, we decided to adhere to the same naming strategy and used the word " leiten" (meaning "lead" and "manage") to denote our controllers.
Common view
const useStore = create<IState>(() => ({ ... })); //clean store without actions
const useController = leiten[Controller](useStore, "dot.nested.path", [options]);
Small Example
Let's create some fake example - load data and then change it.
Pure zustand
const useStore = create<IStore>((set, get) => ({
loadData: async (id: string) => {
try {
set({ loadingData: true })
const response = await getData(id);
set({ data: response })
} catch {
// Todo show error
} finally {
set({ loadingData: false })
}
},
loadingData: false,
data: { user: null, cards: [] },
updateUser: (user: Partial<IUser>) => {
set({ data: { ...get().data, user: { ...get().data?.user, ...user } } })
},
removeCard: (cardId: string) => {
const cards = get().data.cards.filter(card => card.id !== cardId);
set({ data: { ...get().data, cards } })
}
}))
With leiten controllers
import { leitenRequest, leitenRecord, leitenList } from "leiten-zustand"
const useStore = create<IStore>(() => ({
data: { user: null, cards: [] },
}));
// loadingData
const useRequest = leitenRequest(useStore, "data", async (id: string) => getData(id));
// includes update User method
const userController = leitenRecord(useStore, "data.user");
// includes remove Card method
const cardsController = leitenList(useStore, "data.cards", { compare: (a, b) => a.id == b.id });
Using "leiten" controllers empowers you to simplify your state management by removing redundant actions, eliminating unnecessary states, and reducing the complexity of working with nested objects. By adopting "leiten" controllers you can create a cleaner and more streamlined store structure, making it easier to manage and manipulate your application's data.
All actions and states for your zustand store. Examples
- leitenRequest Helps handle requests (promises) and catch errors. Returns a hook with request parameters and provides methods such as action, clear, abort, and set.
- leitenGroupRequest Handles multiple similar requests dynamically. Returns a hook with two overloads and provides methods such as action and clear. Can work with arrays as well as with the normalized list.
- leitenRecord Works with objects and provides methods such as set, patch and clear.
- leitenPrimitive Works with data as if it were a primitive value, but it can be an object, function, or primitives. Provides methods such as set and clear.
- leitenList Works with arrays and provides methods such as set, clear, add, update, remove, toggle, and filter. If the array item is an object, a compare function needs to be set in the controller's options (third parameter).
- leitenModal Helps work with modals and provides a built-in modal manager for cascading modals. Returns hooks with [openState, hiddenState] and provides methods such as open, close and action.
- leitenFilterRequest Same as leitenRequest but provide createFilter and listen methods, which allows you to create an unlimited number of filters for the request. The request will automatically start action when the filter's patch method is called. Or in case listen, the request will be executed if the observed value changes.
- leitenGroupFilterRequest Same as leitenGroupRequest but provide createFilter method, which allows you to create an unlimited number of filters for the request. Works like leitenFilterRequest.
All leitenControllers automatically infer the required types based on the specified path and will throw a TypeScript error if the provided path does not match the controller's requirements or established types.
- Argument of type 'string' is not assignable to parameter of type 'never'.
⚠️ If you encounter an error when specifying the path to your field in the store, it is likely because you are attempting to attach a controller to a field with an incompatible type. Please ensure that the field you are attaching the controller to has a permitted type to resolve this issue.
Library well tree shaking and have dependencies from immer, lodash-es and nanoid
Options
leitenRecord, leitenPrimitive, leitenList have options with callbacks: sideEffect and patchEffect. You can use them to extend basic functionality
const useExampleStore = create<IState>(() => ({ user: null }));
const recordController = leitenRecord(useExampleStore, "user", {
sideEffect: (value: { prev: IUser; next: IUser }) => {
// you can execude here some side actions
},
patchEffect: (value: VALUE) => {
// you can update your entire store here in one tick with value update
},
});
⚠️If leitenList use with regular list (Array) then you should to provide compare function to provide the unique id. For the normalized list (Record) you need to define the getKey function.
leitenRequest and leitenGroupRequest have a useful reactions: fulfilled, rejected, abort, resolved and action
const useExampleStore = create<IState>(() => ({ user: null }));
const recordController = leitenRequest(useExampleStore, "user", async (id: string) => getUser(id), {
fulfilled: ({ previousResult, result, payload }) => {
// do something after successful request
},
rejected: ({ previousResult, error, payload }) => {
// do something after error request
},
abort: ({ previousResult, payload }) => {
// do something after request was aborted
},
resolved: ({ previousResult, payload }) => {
// do something after request was resolved
},
action: ({ previousResult, payload }) => {
// do something before request was called
},
optimisticUpdate: (payload) => {
// if you use this callback, then leitenRequest will automatically switch to optimistic update mode
// the result of this function will be immediately stored in storage and if the promise fails
// it will be rolled back to the previous value
},
initialStatus: ILoadingStatus // initial status if request, 'init' by default
});
FilterRequest
Usage example. There can be an unlimited number of filters. After the patch, inside the lateFilterRequest, a comparison is made with the previous filter value and the request is executed. From one filter you can influence the state of other filters
interface IState {
users: IUser[];
filter: IFilter;
table: ITableFilter;
}
const useExampleStore = create<IState>(() => ({
user: [],
filter: { search: "" },
table: { page: 1 }
}));
const useController = leitenFilterRequest(useExampleStore, "users", async () => {
const props = useExampleStore().getState().filter.search;
return getUser(props)
});
const filter = useController.createFilter("filter", {
sideEffect: () => {
useExampleStore.setState({ table: { page: 1 } })
}
});
const tableFilter = useController.createFilter("table");
useController.listen(useAnotherStore, "period"); // create listener for the external store
// listeners works only if you are use hook returned from leitenFilterRequest in some mounted component on the page
const User = () => {
useEffect(() => {
//initial call
useController.action();
}, [])
return <>
<input onChange={event => filter.patch({ search: event.target.value })} />
<UserTable />
</>
}
Request
All requests working with useLeitenRequestStore. Usually you will never need it, but if you need it, then the record is stored there with all the query parameters. The request key is returned by each leitenRequest
interface IState {
user: IUser | null;
}
const useExampleStore = create<IState>(() => ({
user: null,
}));
const useController = leitenRequest(useExampleStore, "user", getUser);
const User = () => {
// const status = useController(state => state.status) - the same
const status = useLeitenRequestStore(state => state[useController.key].status)
return <>{status}</>
}
leitenMap also can be helpful, example
Group Request
leitenGroupRequest works equally well with both a normalized list and a regular array. If you are using an array, make sure to specify the getKey function, as shown in the example below. Codesandbox link with arrays
interface IStore {
record: Record<string, ICard>,
array: ICard[],
}
const useStore = create<IStore>(() => ({
record: {},
array: []
}));
const useRecordController = leitenGroupRequest(useStore, "record", async (id: string) => getCard(id))
const useArrayController = leitenGroupRequest(useStore, "array", async (id: string) => getCard(id), {
getKey: (value) => value.id
})
leitenGroupRequest return overloaded hook
interface IState {
cards: Record<string, ICard>;
}
const useExampleStore = create<IState>(() => ({
cards: {},
}));
export const useGroupController = leitenGroupRequest(
useExampleStore,
"cards",
async (props: ILeitenGroupRequestParams<string>) => {
return getCard(props.params);
},
);
const status = useGroupController(id, (state) => state.status); //First param is key, better option
// or
const requests = useGroupController((state) => state); // Record with all requests
Feature controller
The features developed via zustand is not modular enough and may have numerous dependencies (import of different zustand stores) that developers may not be aware of. This causes significant problems and confusion when multiple people work on the project's codebase. To provide developers with a way to create independent features and an efficient method for identifying modularity issues, the leitenFeature has been developed. leitenFeature is the Context and automatically Resettable.
Example of a simple feature:
interface IProps {
authorId: string;
}
interface IState {
comments: { id: string; text: string; author: string }[];
}
const [useFeature, FeatureProvider] = leitenFeature(
({ useProps, mount }: IFeatureCreator<IProps>) => {
const useStore = create<IState>(() => ({ comments: [] }));
const request = leitenFilterRequest(useStore, "comments", async (_: void) => {
return [{ id: "1", text: "Hello", author: useProps.getState().authorId }];
});
// works like useEffect
mount(request.action);
// create listener on mount and stop when unmounted
mount(() => request.listen(useProps, "authorId")); // the return value will be called when unmounted
return { useStore };
},
);
The feature code is isolated in its module
Now, all the code required for the feature is within a single scope and isolated
- If a developer sees diverse data and stores within the feature, splitting the feature into two is easier.
- Unwanted dependencies from external stores can be easily identified
One place where dependencies are passed to the feature
All external dependencies required by the feature are defined in the useProps store. This allows
- making informed decisions about the presence of external dependencies
- enables tracking all feature dependencies at once in a single file.
Controller and store names can be simplified
Since the feature has become isolated, it is possible to adopt a system of straightforward variable naming. For example, useStore instead of useAuthorCommentsStore.
Reducing the number of useEffect
Now, inside the feature, mount method is available. Functions can be passed to them to be executed at the corresponding stage of the feature's lifecycle. Method can be called multiple times, creating an array of functions to execute. The passed function can return a function for unsubscribing
In addition, dependency subscriptions can also be handled within LeitenFeature, eliminating the need to move such functionality into any specific component.
Props synchronization on the LeitenFeature side
LeitenFeature takes care of memoization and consolidating all dependencies into one store, freeing the developer from manual dependency management.
Injections instead of singletons provide code reuse safety
If there is a need to use the same code in multiple features, there is no need to use a singleton store and engage in the complex process of reusing a single store for different features.
const [useFeature, FeatureProvider] = leitenFeature(
({}: IFeatureCreator<IProps>) => {
const { useComments, request } = storeCreator();
//some code
return { useComments, ... };
},
);
const [useSecondFeature, SecondFeatureProvider] = leitenFeature(
({ useProps, mount }: IFeatureCreator<IProps>) => {
const { useComments, request } = storeCreator();
//some code
return { useComments, ... };
},
);
const storeCreator = () => {
const useComments = create<IState>(() => ({ comments: [] }));
const request = leitenFilterRequest(useComments, "comments", async (_: void) => {
return [{ id: "1", text: "Hello", author: "" }];
});
return { request, useComments };
};
Now, any store, action, and similar entities can be extracted into a separate function and safely injected into the required feature. In the example, two different features use the same functionality, which they share between themselves.
Tools for identifying architectural errors in code writing
In component code, there should be no imports from other features or singleton stores from previously written features. This way, it can be guaranteed that the component code works only for one feature and operates independently of the environment.
const Component = (() => {
const { useComments } = useFeature();
const comments = useComments(state => state.comments); // right usage
const user = useUser((s) => s.user); // incrorrect, extrenal store used
return <span>{}</span>;
});
A wrapper component for your feature is needed
Feature - it is advisable to make it a memoized component. Inside, it contains the code of your feature. All dependencies required by the feature must be passed in the value
<FeatureProvider value={{ authorId: "name" }}>
<Feature />
</FeatureProvider>
Interaction between features (Advanced usages)
Since there are no global stores now, the question arises of how to implement interaction between nested features. This can be achieved through external feature management. For external management, a third parameter is required, which is exported by leitenFeature.
const [useUnit, UnitProvider] = leitenFeature(({}: IFeatureCreator<IProps>) => {
const useComments = create<IState>(() => ({ comments: [] }));
const request = leitenFilterRequest(useComments, "comments", async (_: void) => {
return [{ id: "1", text: "Hello", author: "" }];
});
const action = () => {
// code
};
return { useComments, action };
});
const [useParent, ParentProvider] = leitenFeature(({}: IFeatureCreator<IDeps>) => {
// To do this we can write a simple helper
const childDeps = create(() => ({ authorId: "123" }));
const childUnit = useUnit.create(childDeps);
const parentAction = () => {
// code
childUnit.action();
// code
};
return { childUnit };
});
In the example, a feature useUnit is described, and we want to use it in the parent feature useParent. We will use the third parameter createUnit, provided by the child feature, and create childUnit inside the parent feature. Now, within the parent feature, we have access to all the functionality of childUnit, and we can use it.
Now, we need to make changes to the code of our provider.
export const ParentUnit = () => {
const { childUnit } = useParent();
return (
<>
<UnitProvider.Unit value={childUnit}>
<Feature />
</UnitProvider.Unit>
</>
);
};
To use our childUnit, you need to pass it to the provider, updating its record beforehand and adding .Unit. Now, our feature has become managed and will work in the context created for it by the parent feature.
In most cases, creating managed features is not required, but this option is provided if more complex interactions need to be established.
Using static stores and methods
Each leitenFeature can have not only a set of dynamic data but also static data. Static data can be used as shared data for all instances of leitenFeature (for example, a modal window or a common data bus). Static data is needed to be used for synchronization within all instances of leitenFeature or outside the leitenFeature context (use only when absolutely necessary, do not abuse).
For using the static part of leitenFeature you should pass two functions into leitenFeature. The first one is the static creator. The return value from this function is available as the second prop of the dynamic creator function.
const [useUnit, UnitProvider, useStatic] = leitenFeature(
({ mount }) => { // mount call once when the first instance of the feature was created
const useCommonStore = create<{ common: string }>(() => ({ common: "" }));
const action = () => {
/* code */
};
return { useCommonStore, action };
},
({ useProps }: IFeatureCreator<IProps>, { useCommonStore }) => { // second prop is the return value from static creator
const useComments = create<IState>(() => ({ comments: [] }));
const request = leitenFilterRequest(useComments, "comments", async (_: void) => {
return [{ id: "1", text: "Hello", author: "" }];
});
const action = () => {
/* code */
};
return { useComments, action, useProps, request };
},
);
Now you are able to use the static object of your leitenFeature. You can use static code out of Provider of your feature. This is the same as regular global stores in zustand, but now you know which feature they belong to and you won’t be mistaken when using them.
const { action, useCommonStore } = useStatic();
Pros and cons
:plus:
- The feature code is isolated in its own module.
- Dependencies are passed to the feature in one place.
- Controller and store names can be simplified.
- Reducing the number of useEffect.
- Props synchronization on the LeitenFeature side.
- Injections instead of singletons provide code reuse safety.
- Easily test the business logic of a feature
- Tools for identifying architectural errors in code writing.
- No circular dependencies.
- No global stores. But if necessary, possible to create a global store.
:minus:
- It is necessary to wrap the feature with a provider.
- Limited interaction between different features.